Hermes Labs
Find and fix silent failures in production AI systems
Visit hermes-labs.ai ↗What is Hermes Labs?
Hermes Labs provides agentic infrastructure and reliability engineering for autonomous AI systems. The platform detects failures that standard evaluations miss—such as guardrails dropping mid-conversation, fabricated tool calls, and memory degradation—then engineers fixes with verifiable evidence. The team has contributed four merged fixes to LangChain, Microsoft Semantic Kernel, and DSPy, plus maintains an open-source catalog of 19 tools.
The company offers research papers on AI failure modes, scoping calls for production issues, and open-source utilities like Fidelis Memory, which preserves agent context verbatim instead of paraphrasing. Fidelis uses local-first retrieval (BM25, dense vectors, and reciprocal-rank-fusion) and integrates with Claude Code and other agents.
Momentum
Daily activity over the last 30 days: upvotes, unique visits, new GitHub stars and Hacker News points.
How do we know it's vibe coded?
vibetrend discovered Hermes Labs automatically. Evidence: GitHub hermes-labs-ai/fidelis; repo has GEMINI.md. View source.
Similar vibe-coded apps
All-in-one agent platform connecting global SaaS and China enterprise systems
Pre-action firewall for AI coding agents that blocks mistakes before execution
Open source AI gateway routing across 350+ providers through one endpoint
Open-source AI job search agent that runs locally in your CLI
A second brain for the AI era. Markdown notes with Git and AI agents.
Local code graph for AI-assisted reviews with 63x context reduction
AI-native engineering philosophy and plugin for coding agents
Open-source framework for building autonomous agents with purpose-built UIs